Debt Megastructure

Benjamin Bratton describes The Stack as an accidental megastructure, and if it is that, it is also an accidental data structure, and an accidental planetary financial instrument.

User – The Illiquidity of the Social

Social media currencies (part of the broader Bratton term fabricated currencies) are a recurrent pattern in the platforms of our era: likes, shares, citations, retweets, +1. From a user finance point of view, one of the striking things about a social media currency is how hard it is to spend. Certainly it comes with attention, you can earn money off it as a resource, and collections of it have compounding, snowball effects. Yet the mechanics are strikingly indirect for objects which were digital from the start of their short history. You can’t buy a sandwich with Facebook likes the way you can with dollars. The likes are instead at the start of a multi-step supply chain of, say, social media capital, advertisement placement, rent through advertisement clicks on the social media asset, delivery in a traditional exchange currency, and finally use at the sandwich shop counter. Reversing the direction, entire identities can be constructed and sold en bloc, likes can be manufactured out of human or robot labour, but it’s a logistical chain, not a retail transaction. As assets go, social media capital is rather illiquid.

Actual exchange currencies created within platforms that happen to be digital, like the virtual gold in the game World of Warcraft, are distinct from this. They just behave similarly to traditional exchange currencies, including ease of trading into the exchange currencies of commodity money, state-backed fiat currency, bitcoin, and so on.

Social media currencies are numeric, constantly accumulating integer counts, but that’s their most obvious functional similarity to currencies of exchange. Counts are accumulated around nodes in a graph, not consolidated into aggregated scalars. So I might have 10 paltry followers on Twitter, but also one very popular tweet about whales that was retweeted hundreds or thousands of times. Though the two aspects interact – I may pick up new followers in the wake of such a flash of cetacean insight – the platform itself doesn’t add the two together, or combine them according to any more sophisticated formula. The record of social credit is left in the form of a graph, because it’s perfectly traversable algorithmically, and that’s the form in which it’s most useful to the corporations and users that run and support it. The major platforms don’t try to track or represent the graph across platforms, either. There’s little advantage to Facebook drawing attention to how popular you are elsewhere, at least for now.

Though summarizing the graph as a single number has been avoided, Mark Zuckerberg did, until 2015, promote the idea of a single identity linked to a physical body and political identity, even saying not using a real name on Facebook showed a “lack of integrity”. A single identity increases the power of the Facebook social graph by linking nodes. At the end of 2015, Facebook stepped back from the policy, under pressure from civil liberties organizations. Perhaps its executives also realized related identities were unlikely to stay in disjoint parts of a social graph for long. It certainly hasn’t stopped the construction of Facebook’s comprehensive advertisement-surveillance infrastructure, much of which is devoted to constructing links back to a common identity.


Interface – The Means of User Production

The two groups most publicly eager to consolidate social presence into a single score with direct material rewards have so far been marketing consultancies and the Chinese Communist Party. On the marketing side, Klout offers a consolidated score of social media influence with the prospect of free stuff for high score individuals with the right consumer demographic profile. Meanwhile, the Chinese government is building vast databases for all citizens to calculate a similar score, also including financial credit, traffic violations, schools, industrial accident safety, online rumourmongering, civil service performance, and so on. Alibaba’s Sesame credit 芝麻信用, a publicly viewable credit rating incorporating other trust indicators, is a prototype, and dating site Baihe 百合网 shows that score on user profiles.

The CCP policy documents so far available are more concerned with comprehensive coverage than explicitly defining a single personal score. Its also worth keeping in mind that such documents can be more akin to discussion papers or think tank proposals than settled policy. Firm plans or not, they anyway combine with the already live prototypes to give ideas of one future system. Though we could take the primary goal as control, the goal actually stated is:

[E]stablishing the idea of an sincerity culture, and carrying forward sincerity and traditional virtues[.]


– Planning Outline For The Construction of a Social Credit System (intro)

Achieving this cultural gardening through construction of software systems implies a platform view: one that recognizes, like Bratton, that “computation is a logic of culture”. The CCP’s “sincerity culture” could also be translated “integrity culture” or “honesty culture” and be more in keeping with liberal political norms. The policy document goes on with a long enumeration of parts of society that should be improved by such a system, across all layers of the social stack, from agriculture through to usernames, and organized more or less by the structure of government departments. Corruption in government itself is also a target:

Accelerating the construction of government affairs sincerity


– Ibid, II (1)

The Propeller Group noted this intersection between modern consumer advertising and modern consumer communism in their 2011 art piece Everyone Is Equal, where they commissioned an advertising agency to rebrand the Vietnamese ruling party’s ideology. An aggregated social media score is more of a statement that Everyone Is Arithmetically Comparable, which is just as universally radical, if rather less egalitarian.

Aligning public form with civic virtue, especially in the context of traditional values, is a Confucian concern, that suggests the rectification of names, or in a more imperial mode, Legalist unification of rewards and punishments. Construction of virtuous culture through government platforms and public integrity is hardly a uniquely Chinese concern, however: it’s found in Leviathan, Gulliver’s Travels and Scooby Doo. Conceptually rhyming with all three, Klout entreats us to “Be Known For What You Love”. “Accelerate Sincerity”, with its dynamic notes of entrepreneurial self-actualization, could easily substitute, if needed.

Sincerity, like all interface relationships, constrains in two directions. Outer appearances more closely reflect inner thoughts. Inner thoughts are trained to conform more to outer appearances, and their constraints.


Address – Short Term Memory

A consolidated score is as much an input to human cognitive processing as to algorithmic calculation. Though an algorithm can use a single numeric input, it can also use the richer parameters of multiple social media graphs. The design constraint is short term conscious human cognition, which struggles with vector arithmetic, but can certainly understand if their personal score is five points lower than their neighbour’s.

In both cases the goal is to have an individual user optimize their representation as a product. In one case it’s a consumer product, in another a loyal product of the State, but one can blend easily enough into the other. From each according to their volubility, to each according to their brand, perhaps.

A consolidated single number is the form Cory Doctorow chose for whuffie, the reputational currency in his 2003 novel Down and Out In The Magic Kingdom. Whuffie anticipates both Klout and communist sincerity scores, and is already a fusion of the two. In Doctorow’s utopia, Californian in outlook if not strictly location, neural lace implants are ubiquitous, everyone is always online, and checking whuffie is as routine a part of meeting people as waving hello. There are more complex structures feeding the algorithmic calculation of whuffie across all the people in the world you might be associated with, but there is always a headline number which characters react to as a first impression. Using a consolidated score not only simplifies cognitive and explanatory load for the characters in the world of the book, it also simplifies it for the reader, becoming a shorthand for conveying a character’s social fortunes. Thomas Piketty notes that in the stable price environment of 19th century Britain and France, novelists were able to make use of net worth in a similar way, as a rapid quantitative social shorthand. Piketty goes on to use literary criticism to explore the economic history of inequality; Doctorow uses a quantitative social shorthand to write a novel exploring the the economic future of inequality.

A future where your success in life depends on generating social media tokens of approval can also sound like a hellish dystopia (and indeed there’s already a Black Mirror episode with exactly that premise). Swapping an exchange currency for a social currency doesn’t get rid of misery, as Doctorow’s novel acknowledges right in the title. If misery is moderated, it is by other means, such as the universal entitlement to basic food and shelter in Doctorow’s Bitchun’ Society.

The implied scale and sophistication of whuffie infrastructure is vast, though not that much larger than the online social networks that now exist in 2017. The mechanics described in the book are little more sophisticated than tracking likes and content selection in existing social media feeds. Facebook counts 1.9 billion active users, so as a technical scaling problem they have less than one more order of magnitude to go.

An aggregated score is most useful to memory-constrained users of the platform. As in Bratton’s Stack, User is a role not limited to humans, but could be an organization, a program, a houseplant, or a househusband. These users could be phone apps with strict latency ceilings doing basic comparison checks, analytics needing first order filters for cutting down huge data sets to manageable size, or humans quickly sizing up someone they just met.

The existence of whuffie in this consolidated form still implies a social debt megastructure calculating it. Unlike exchange currency, whuffie isn’t in your wallet or decentralized bank account, but calculated in the cloud, centrally accounted for by the platform that issues it.

In fractional reserve banking, private banks themselves create money by issuing loans only partly backed by current deposits. Banks balance flows of money as a dam leaning against a reservoir of capital, and if they get the flows of loans and repayments wrong, they collapse. In the big social media platforms of today, users create new tokens of currency every time they like and share. In early experiments in scaling social media circa 2000, like the first moderation system on news and commentary site Slashdot, moderation points were rare and allocated by the platform in strictly limited tranches. On the Twitter or Facebook of today, everyone is a moderator all of the time, and so everyone is also a user-sized fractional reserve microbank of social media currency. These reputational currencies are both sincerity currencies and (reversibly) duplicity currencies, like reputation down the ages, but digitally codified in data structures and circulated by algorithmic feeds.

We’ve learned, by now, that social media is just as potent a channel for emotional engagement as mass media, and it may be that the sense of discovery from refreshing a feed or following a link deepens the attachment to the discovered information. Once the connection is made, the adaptive network of the whuffie bank acts as a sincerity amplification machine. In the case of fake news, it’s a duplicity amplification machine.

As with exchange currencies, issuing more tokens is an inflationary process that devalues each token and increases the supply available for interaction. Inflation is a form of spam. Every participant in a reputation economy has to manage their signal to noise ratio to maintain participation without devaluing their tokens by wasting other users’ time. A like is a personal bitcoin mined with proof of charm, Facebook as the only validator, that you can only spend once. Likewise a social network with an established userbase, but falling participation, is in a liquidity trap.

Currency tokens are a form of metadata on top of the underlying value in the system. In social networks we would usually call this “content”; in economics, “wealth”, or “capital”.


Cloud  – Network Shadows

Facebook, as the platform for issuing a major reputational currency, is both a reputational merchant bank and a central bank. It and Twitter are also ontology casinos, revealing idea reinforcements and reactions in unpredictable, addictive ways. The short tactile feedback cycle of checking for feed updates on your phone is the same physical action-mechanical response-decision loop as a slot machine.

Facebook scientists have already shown they can subtly and measurably alter the emotional response of users based on the content of their newsfeed. This is not a terribly surprising or new result: only the algorithmic lever is new. Mark Zuckerberg has become the Governor of an accidental Reserve Bank of Sincerity.

State reserve bankers won’t shut up about money, in particular prices. They even have a jargon – Open Mouth Operations – for how talking about prices publicly can measurably move the market. Accidental Reserve Bank Governor Zuckerberg goes on tours of fifty US states where he guts fish and eats in diners like a junior politician running for office. He talks about money too – he’s still a CEO –  but only about advertising rates and Facebook profits.

Social media companies trade our reputations in dark pools. The links between monetary prices and social media currencies are concealed within the platform by design. Two interfaces exist for two types of user – reputational users, and trading users. The trading interface is by and large just another market.

The reputational user coins their own currency, but exists in a space constructed with systematically occluded prices. This experience contrasts with the neoliberal experience of ever-expanding monetization – of the household, of schools, and so on. In one sense it is localized whuffie consumer communism. In another it is a cattle yard where the user-commodity is not an agent in the market. Of course a cow can’t buy a cow, whereas a person or a corporation can buy advertisements, even advertisements about themselves. Note there are still immediate interface-imposed limits. For example, you can’t buy your own advertisement impressions, or yourself as a customer (for instance, as a way of eliminating advertising). You can block or ignore advertisements, but not purchase your way out of platform bondage, only exit the platform entirely. You can’t close the loop, except accidentally, or partially. The platform holds information about your own mental state and value that you do not. It has a shadow copy of yourself, a model of your profile and behaviour, which it rents to advertisers. Trying to close the loop by understanding the model, like watching ourselves through a drone camera, is a common interaction in the Stack, “first person access to third-person experiences of first-person experiences”, as Bratton describes both the novel A Scanner Darkly and this specific design affordance. With a social media platform, however, the lens lets us make out a grimy first-person plural at best. For commercial, privacy, and usability reasons, the platform forcefully decouples self-as-product and self-as-trader. This muddy scanner, or shadow copy, platform design is a software cousin of Bratton’s glass fort, a critical term for built public architecture which uses expressionist flourishes and extensive exterior windows to ensure both security and impressions of transparency.

CalTrans District 7 HQ, Thom Mayne (Morphosis). Wikicommons.

The shadow copy is not a high fidelity one. It’s more of a software voodoo doll than a deep clone. It does not need to map each identity perfectly to a single corporeal self. Not having personal access to the specific proprietary codebases in question, it may not even be an identifiable singular data structure, but rather an emergent informational effect. The functional behaviour is similar either way, though perhaps more clearly indexable shadow identities are easier to sell.


Platform Contingency

In “The Ontology of Finance: Price, Power and the Arkéderivative”, Suhail Malik constructs a theory of finance as a sovereign power, particularly the ways derivative markets internalize time and contingency in data structures within the overall system of capitalist prices. Malik uses, amongst many things, Ayache’s idea of a medium of contingency. For Malik, “finance theory is a variant of media studies”, and both finance and the state are mediums of contingency.

[E]very price is intrinsically an occasion of political economy, of what, where and how much power over futurity is to be had.

  – Malik, “Ontology of Finance”

Malik does not describe finance as sovereign but finance-power, structured by prices, competing with or entwined with state sovereignty. Bratton states at the outset of his book that the Stack is another model of sovereignty set in contrast to Westphalian territoriality. Like states, platforms too are a medium of contingency. They define what is impossible, and what are spaces of empowered possibility, in code.

Interfaces are reversible, so there is a channel by which reputational currencies impact prices and help define the platform. Social media currencies are in a loose sense derivative on the content they are anchored to. But they are not as precise or diverse as financial derivatives, not as all-encompassing, and lack the theoretical instrumentation of Modern Monetary Theory, or the Black-Scholes-Merton model. Every like is an occasion of political economy, but a deflected and distracted occasion. For now, these social media spaces are distinguished from finance-power and platform sovereignty only in that they demonstrate a demonetized, reputational currency space, not that they weaken platform sovereignty or finance-power in any way. Quite the opposite: otherwise they wouldn’t exist.

The places where reputational currencies begin to shape the platform as a media of contingency are social media features on sites for developers themselves. On github, which hosts open source code, a project owner votes +1 to accept a submitted patch. Patches are fragments of changing contingency and the atoms of platform sovereignty. On StackOverflow, a site for expert discussion of very specific programming problems, reputational currencies reinforce good solutions, when users vote up good questions and good answers. A very well-framed question can even trigger, socially, design fixes in the software it critiques, just through the clarity of showing a concrete problem. (Improvement through humans expressing machinic sincerity.)

The scope of these social media effects is determined by the openness of the code concerned. Facebook and Google are private firms, and though they do contribute to open source, much of the code that defines their platforms is private. The ability to submit or approve patches defines the two elite classes of platform social media User. What code actually gets built and run at a private firm is defined by capital, making a social media company a site of both platform sovereignty and finance-power. Non-elite users, even whuffie billionaires, are inputs and commodities, not sovereign agents.

These power-disparities could deepen and entrench, or be obverted by new or mutated links between reputational currencies and platform sovereignties. And there are state sovereigns yet. The US antitrust regulator could force Facebook to fork. Startup hedge funds with Ethereum constitutions could use shareholders votes in reputational currency to confirm their legal officers. Nauru could be merged with Amazon in return for aggressive decarbonization of the world logistics chain. The Chinese Communist Party could open up affiliate membership worldwide and distribute bitcoin dividends to everyone quantifiably sincere. The Stack is a computational planet-system terraforming itself. Managing it is absurd, and changing it happens everyday.


Ayache, E. (2011). The medium of contingency. Pli.
Bratton, B. H. (2016). The Stack: On software and sovereignty. MIT Press.
Brooker, Charlie (Writer) and Wright, Joe (Director), (2016, Oct 21). Nosedive [TV series episode]. In Black Mirror
Dick, P. K. (2011). A scanner darkly. Houghton Mifflin Harcourt.
Doctorow, C. (2003). Down and out in the magic kingdom. Macmillan.
Kramer, A. D., Guillory, J. E., & Hancock, J. T. (2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences, 111(24), 8788-8790.
Lanchester, John (2017, August 17). You Are The Product, London Review of Books.
Lee, Dave (2015, December 15). Facebook amends ‘real name’ policy after protests, BBC,
Malik, S. (2014). The ontology of finance: Price, power, and the Arkhederivative. In Collapse Vol. VIII: Casino Real (pp. 629-811). Falmouth, UK: Urbanomic.
People’s Republic of China State Council, Rogier Creemers (trans), (2014), 国务院关于印发社会信用体系建设 “Planning Outline For The Construction of a Social Credit System”,
Piketty, Thomas. Capital in the 21st Century. Cambridge: Harvard Uni (2014).
Propeller Group, Everyone is Equal, (2011),

Slot machine connection to design tactility of social media noted by Chenoe Hart and “Memento Hikikomori” on Twitter

2 thoughts on “Debt Megastructure

  1. Pingback: Carbon Refactoring | Conflated Automatons

  2. Pingback: Retooling the Social Media Contract | Conflated Automatons

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.